It was only after 10 years as a specialist treating anxiety disorders that I finally realized the central role of runaway positive feedback. The insight was hugely helpful for me and my patients. Since then, my thinking has expanded, even snowballed!, to consider the role of vicious cycles in many other diseases. The literature has many great specific examples, but no comprehensive review article. I will write one eventually. In the meanwhile, this Nessay offers examples of how runaway positive feedback loops cause disease in hopes of generating interest, comments, and perhaps enough shares to illustrate the positive side of positive feedback!

The standard metaphor for runaway positive feedback is a snowball rolling downhill. With each turn it gets larger and rolls faster, becoming larger and faster yet. Real snowballs almost always collapse as they are getting started, but AI can dramatize the metaphor. Another familiar example is the shriek from a public address system when the microphone feeds output from the speaker back into an amplifier in a vicious cycle that stops only when the mic is moved away from the speakers.

Negative feedback is the opposite of positive feedback, it dampens extremes and stabilizes systems in the process called homeostasis. Negative feedback is not all good and positive feedback is not all bad! In fact most control systems involve positive as well as negative feedback loops. However, those details are for another article, the thesis here is simply that runaway positive feedback is an important cause of disease that deserves much wider recognition.

Disease from runaway positive feedback

We start with the heart

When a tiny flap of atherosclerotic plaque loosens in a coronary artery, it creates turbulence that sets off clot formation, more turbulence, and more clotting in a positive feedback loop that blocks the artery and causes a heart attack.

Recognition of the role of positive feedback in cardiac failure is now improving clinical care. Insufficient cardiac output, for any number of reasons, causes vasoconstriction and fluid retention that further compromises output in a vicious cycle.

Subscribe now

Inflammation and immune responses

These defensive responses are intrinsically vulnerable to runaway pathology because inflammation can cause damage that causes more inflammation, because their regulation mechanisms usually include an initial positive feedback boost to get them up and running fast, and, as Steve Frank has noted, because intense competition with pathogens creates peaked fitness functions that make even optimal responses costly.

The scratch—itch cycle is common and often hard to control.
Skin damage → inflammation → itching → scratching that damages skin more.

Cytokine storm is far more serious. Cell damage releases cytokines that can create more damage and more cytokine release in a vicious cycle that was responsible for many deaths in COVID-19. Most studies using steroids to block inflammation still assume everyone is the same, but finding ways to distinguish patients with excess vs. deficient immune responses would save lives.

Alzheimer’s disease is a product of useful defenses gone awry. Amyloid-beta is an antimicrobial that causes inflammation that can cause further damage and abnormal protein folding in a runaway process that damages synapses then kills neurons.

Share Nessays

Parkinson’s Disease is common. We are all vulnerable because dopamine (DA) is metabolized by monoamine oxidase to 3,4-Dihydroxyphenylacetaldehyde (DOPAL), an aldehyde that damages DA storage vesicles, releasing more DA that is metabolized to more DOPAL, creating more damage.

Appendicitis is an indirect product of inflammation. Initial inflammation at the neck of the appendix compromises its blood supply, allowing bacteria to grow, causing more inflammation, a further reduced blood supply, until growing bacteria burst the appendix.

Systemic lupus erythematosus(SLE) is the product of multiple interacting positive feedback loops that create antibodies to an individual’s DNA. Failure to fully clean up after normal cell death exposes DNA to B cells that make antibodies that bind them and create immune complexes that create inflammation in small vessels in the kidneys, skin, joints and the heart, releasing complement that causes further damage and DNA exposure.

Behavioral and emotional disorders

Anxiety disorders was my starting point. Positive feedback is central to their pathophysiology.

Performance anxiety is mostly a product of vicious cycles. Fear of being able to urinate in a public bathroom, causes inhibition of the parasympathetic system, making urination impossible. Fear of being impotent is self-reinforcing by the same route. Fear of speaking in public, can cause sweating, shaking and voice quivering that disrupt performance and make the next attempt more difficult yet. Sharing these principles is essential for helping patients to understand how behavior therapy works to disrupt vicious cycles and why exposure therapy is essential, despite being difficult and anxiety-provoking.

Panic disorder often results from fear of fear. Even a slight increase in heart rate can make a person anxious about the possibility that a panic attack is coming, the anxiety makes heart rate go up further and creates additional symptoms that escalate into a full-blown panic attack. The smoke detector principle explains why false alarms in the panic control system are common and normal; when a danger might or might not be actually present, then expressing the protective response can be far less costly than the risk of catastrophic outcome if it is not expressed. Such signal detection analysis has yet to be applied routinely to other physiological defensive responses.

Addiction is the straightforward outcome of positive feedback. Drugs that hijack learning mechanisms cause increased drug taking and further brain changes. Withdrawal symptoms are also relevant, but probably less so than the simple effect of drugs on learning mechanisms.

Eating disorders are a tragic example. It seems so straightforward to simply stop eating for a few days to lose weight, but such a restriction triggers famine protection mechanisms that result in uncontrolled gorging, causing increased fear of gaining weight, resetting the body mass set point higher, and subsequent new resolve to more strongly control eating in a vicious cycle that is all too often fatal.

Bipolar disorder results from mood dyscontrol and positive feedback. Cyclothymic individuals have high gain in their mood control system causing fast extreme swings. When the gain is pushed higher yet, the system goes to extreme highs or lows and gets stuck there, in the characteristic control system mode of bistability that characterizes bipolar disorder.

Implications

Taken together, these examples have substantial theoretical implications for how we understand disease. They span diverse organ systems and pathophysiological mechanisms, yet share a common architecture: perturbed protective or regulatory responses amplify rather than dampen dysfunction. Recognizing positive feedback as a fundamental disease mechanism—not just a feature of specific disorders—could transform how we approach research, prevention, and treatment design. The question isn't whether positive feedback matters in disease, but how many conditions we've misunderstood by failing to recognize it.

Recognizing the role of vicious cycles in disease also has practical implications. Positive feedback loops are best disrupted early, before they gain momentum or by imposing powerful negative feedback. Beta-blockers in heart failure interrupt sympathetic amplification. Cognitive therapy for anxiety works by repeatedly exposing patients to feared situations without catastrophic outcomes, weakening the fear-symptom connection. For patients, understanding feedback mechanisms can be therapeutic—recognizing a vicious cycle often helps to disrupt it. For researchers, recognizing such feedback loops is the crucial first step towards finding strategies for disrupting them.

Many more examples are available, but I'll save them for a longer article in a journal. In the meanwhile, I'd love to hear from people who are interested in the role of control systems and positive feedback in causing disease.

Leave a comment

PDF versions of Nessays are available at https://www.randolphnesse.com/nessays

References for additional reading are below

• Blithikioti C, Fried EI, Albanese E et al. Reevaluating the brain disease model of addiction. The Lancet Psychiatry 2025;12:469–74.

• Frank SA. Disease from opposing forces in regulatory control. Evolution, Medicine, and Public Health 2023;11:348–52.

• Goldstein, D. S. (2020). “Sick-but-not-dead”: Multiple paths to catecholamine deficiency in Lewy body diseases. Stress, 23(6), 633–637. https://doi.org/10.1080/10253890.2020.1765158

• Goldstein J. Attractors and Nonlinear Dynamical Systems. 2011.

• Kitano H. Violations of robustness trade‐offs. Molecular Systems Biology 2010;6:384.

• Mackey MC, Glass L. Oscillation and Chaos in Physiological Control Systems. Science 1977;197:287–9.

• Nesse, R. M., Labov, J. B., & Madhavan, G. (2025). Explanations for failures in designed and evolved systems. PNAS Nexus, 4(4), pgaf086. https://doi.org/10.1093/pnasnexus/pgaf086

• Nesse, R. M. (2023). Evolutionary psychiatry: Foundations, progress and challenges. World Psychiatry, 22(2), 177–202. https://doi.org/10.1002/wps.21072

• Nunes A, Singh S, Allman J et al. A critical evaluation of dynamical systems models of bipolar disorder. Transl Psychiatry 2022;12:416.

• Packer M. The neurohormonal hypothesis: A theory to explain the mechanism of disease progression in heart failure. Journal of the American College of Cardiology 1992;20:248–54.

• Shlomchik MJ, Craft JE, Mamula MJ. From T to B and back again: positive feedback in systemic autoimmune disease. Nat Rev Immunol 2001;1:147–53.

• Tang L, Yin Z, Hu Y et al. Controlling Cytokine Storm Is Vital in COVID-19. Front Immunol 2020;11:570993.

It would be wonderful if genetic information could predict complex phenotypes accurately, but that looks increasingly unlikely. Missing heritability is substantial, it won’t be explained by newly discovered rare variants [1], and the predictive power of polygenic scores is limited and lower yet in new populations [2].

These limits arise because the path from genes to traits is rarely like the “genes for” model that naively assumes that specific genes code for specific traits. Instead, genes and regulatory regions interact with each other and gene products to create winding paths through a tangled bank of organically complex systems. As Mackay and Arnold dryly conclude in a 2024 review, “relationships between genotypes and phenotypes are often more complex than can be detected with the methods usually employed [3].”

Fred Nijhout and I are working to see if analyzing interaction effects that emerge from a mathematical model of one-carbon metabolism can identify emergent epistasis that could better describe gene → phenotype pathways. A foundation for that project is systematic consideration of the many ways that genes can influence phenotypes. Here is a brief summary.

Subscribe now

The “Gene for” model, in which variation in a gene induces variation in a protein that causes variation in a trait, is often applied inappropriately to complex traits.

Most traits are polygenic, that is, they are influenced by many genes. For instance, over 12,000 genetic variations influence human height. Together they account for only half of its heritability. What accounts for the rest?

Complexity arises not only because many genes influence one trait, but also because most genes influence many traits. This is pleiotropy. It is especially evident in the findings from studies of mental disorders; most genes that increase the risk of one disorder also increase the risk of others, dashing hopes of finding specific genotypes that could define specific disorders and resolve the confusion that swirls around psychiatric diagnosis.

Now it gets more complex fast. As Bateson and Fisher noted nearly century ago, genes influence the effects of other genes, some “standing upon” (stasis) or otherwise influencing the effects of others in ways that create nonlinear epistatic interactions.

Genetic epistasis creates plenty of complexity, but it is augmented by physiological epistasis that arises from nonlinear interactions of genes and enzymes whose activity levels influence gene expression and the activity levels of other enzymes in complex metabolic systems with feedback loops and nonlinear kinetics that interact with each other environments to influence traits.

Finally, we come to the Convoluted Complexity in Organic Systems (CCOS). Its reality is unwelcome, to say the least. It severely limits hopes that genetic information will ever be able to reliably predict disease risk and other complex traits. This may reassure those to those who hate the idea that genes are destiny but it should dismay all of us who had hoped that genetic information would allow diseases to be reliably anticipated and prevented. For prospective parents who are thinking of purchasing a service that sequences DNA from fertilized eggs to enable personalized eugenics, the reality of CCOS encourages rethinking the whole idea.

The conclusion is simple: the pathways from genes to traits are a maze in a tangled bank of organic complexity that fundamentally limit the power of genetic predictions about complex traits. The larger implication is profound: the structure of complexity in organic and designed systems is fundamentally different. Failure to recognize the differences encourages tacit creationism that obstructs progress.

Share

Leave a comment

Subscribe now

1. Wainschtein, P., Zhang, Y., Schwartzentruber, J., Kassam, I., Sidorenko, J., Fiziev, P. P., Wang, H., McRae, J., Border, R., Zaitlen, N., Sankararaman, S., Goddard, M. E., Zeng, J., Visscher, P. M., Farh, K. K.-H., & Yengo, L. (2025). Estimation and mapping of the missing heritability of human phenotypes. Nature. https://doi.org/10.1038/s41586-025-09720-6

2. Kullo, I. J. (2025). Clinical use of polygenic risk scores: Current status, barriers and future directions. Nature Reviews Genetics. https://doi.org/10.1038/s41576-025-00900-8

3. Mackay, T. F. C., & Anholt, R. R. H. (2024). Pleiotropy, epistasis and the genetic architecture of quantitative traits. Nature Reviews Genetics, 25(9), 639–657. https://doi.org/10.1038/s41576-024-00711-3

Research on mental disorders has been guided by hopes of finding specific causes that define specific disorders. That strategy worked for the rest of medicine, so it was enthusiastically adopted by psychiatry. The descriptions of mental disorders in the 1980 DSM-III were viewed as placeholders until research could discover the specific causes that would define specific disorders. Four decades of intensive research later, it is increasingly clear that psychiatry’s Holy Grail does not exist.

Hopes of finding specific genes that cause specific disorders motivated studies that found thousands of relevant variants, but they each have only tiny effects and most influence many disorders. Hopes that specific disorders resulted from excesses or deficits of specific neurotransmitters or receptors were well worth pursuing, but they too have not been fulfilled. Hopes of finding abnormalities at specific brain loci have spurred development of incredibly sophisticated methods that have revealed unimagined details about brain structure and function, but no specific brain abnormalities that cause specific disorders. It is time to accept the evidence: what we have been searching for does not exist in anything like the simple form we hoped for.

The disappointment is crushing. Mental disorders are a plague on our species, solutions are needed, and searching for specific causes is undeniably worthwhile, but it is time to respect the data. Most mental disorders don’t have specific causes. They are, like most other medical disorders, failure modes in functional systems that can have multiple causes. Anemia, rashes, and chronic cough can have many causes, but usually a specific cause can be found in a specific individual. Congestive heart failure and epilepsy are, in contrast, failure modes in functional systems that often result from multiple interacting causes. Like many other medical disorders, mental disorders are syndromes whose consistency comes not from specific causes but instead from failures in adaptive systems that can have many causes.

This evolutionary approach is not an alternative to looking for factors that explain why some individuals get a disorder, it is an additional perspective that addresses a different question: why did natural selection leave the system vulnerable to this kind of failure? There are several possible answers. Some apparent disorders are actually adaptive responses; examples include vomiting, fever, anxiety, and depression. Such responses seem pathological because they are aversive, but they can go too far to cause serious conditions such as dehydration, seizures, and anxiety and mood disorders. Other disorders, such as obesity and myopia, are products of bodies mismatched with environments far different from those humans evolved in. Natural selection pushes some traits to a performance peak near a cliff edge, so that inevitable variation leaves some individuals off the edge and vulnerable to disease. Autoimmune disorders are a good example, but the extreme ambition associated with some mood and psychotic disorders may also be relevant. Many problems result when a wrenching transition to a new niche creates strong selection that improves adaptation overall at the cost of disrupting existing systems. Shifting to a bipedal stance resulted in hernias, knee and back pain and varicose veins that persist after two million years of subsequent evolution. The shift to the social/cultural/language niche also created strong selection with net benefits but also costs, perhaps including vulnerability to the kind of system failure we call schizophrenia, a hypothesis supported by data showing that newer alleles that influence schizophrenia tend to reduce the risk. Control systems in general are vulnerable to failure, especially in modern environments, as sadly illustrated by the prevalence of eating and substance use disorders.

I will return in future essays to strategies for analyzing mental disorders as products of adaptive systems that natural selection has left vulnerable to failure, but today’s conclusion is simply that the search for specific causes the define specific mental disorders was a good idea that has not panned out. It now needs to be augmented by an evolutionary medical analysis of disorders as failure modes in adaptive systems. In the meanwhile, an extensive technical treatment of these issues is available in an article in World Psychiatry, and the challenges of creating a genuinely medical foundation for psychiatric nosology are covered in an article with the great and wonderful psychiatrist/scientist/philosopher/friend Dan Stein, who tragically passed away last week.

Viewing Diseases As if they are Adaptations shaped by natural selection (VDAA) is increasingly recognized as an error, but it remains common. Diseases are not traits shaped by natural selection. They don’t have evolutionary explanations based on their adaptive functions. The correct objects of explanation are, instead, universal traits that make all individuals in a species vulnerable to a disease.

Does the creativity of some individuals who have schizophrenia give selective advantages that explains the persistence of the responsible genes? Can the special abilities of some individuals with autism explain its persistence? Are cystic fibrosis alleles selected for because they prevent fatal diarrhea? Is anorexia nervosa a facultative adaptation that responds to famine by inducing some individuals to run many miles looking for food to benefit the group? Was ADHD selected for because it increases curiosity or creativity? Published articles advocate for each of those hypotheses, but they are inconsistent with evolutionary theory. Genes and traits that reduce average inclusive fitness tend to be selected out, even if they benefit some individuals or groups. With only a few exceptions, uncommon traits are not adaptations. The default evolutionary explanation for diseases is not that they are somehow useful, it is instead that natural selection’s power is limited, and it shapes traits that maximize gene transmission at the expense of robustness

VDAA persists nonetheless, like a mutation that arises again as fast as it is selected out. I think it is a product of deep human cognitive tendencies. When George Williams and I first tackled the mystery of why natural selection leaves organisms vulnerable to disease, we spent hours speculating about ways that atherosclerosis, cancer, and schizophrenia could give advantages. Only after a month of frustrating discussions did we finally recognize our mistake: diseases don’t have direct evolutionary explanations. The correct objects of explanation are, instead, universal traits that make all individuals in a species vulnerable to a disease. Examples include immune cells in the endothelium, the ability of liver cells to multiply, the narrow bh canal, and the body’s limited abilities to prevent mutations, cancer, and infection.

That crucial insight spurred us to describe six possible evolutionary explanations for traits that cause disease vulnerability. The first three reflect the limitations of natural selection:

• it can’t get to or preserve optimal designs ,

• it is too slow to keep up with shifting environments, and

• it is far too slow to keep up with fast-evolving pathogens.

The other three arise instead from the power of natural selection:

• it shapes tradeoffs that compromise individual robustness in the service of increased performance,

• it maximizes reproduction at the expense of health and longevity, and

• it shapes costly painful defensives that impose substantial risks.

Those six categories have been widely used to understand universal traits that make bodies vulnerable to disease. Multiple explanations usually apply, but the tendency to emphasize only one is a topic for a different essay.

Why do efforts to explain diseases as adaptations continue when a framework is available for understanding traits that make a species vulnerable? I see two main possibilities: 1) humans cognition is predisposed to find functions, and 2) many diseases are associated with benefits.

Battles about adaptationism still smoulder despite consuming gigawatts of emotion and thinking. Gould and Lewontin’s 1979 spandrels article attacked speculations about adaptations effectively, but it said little about how to frame and test hypotheses. Forty-four years later, the phrase “just-so-stories” still suffices as a critique in some circles, as if the fields of evolutionary physiology and behavioral ecology did not exist. Such global criticism does little to address a problem that remains real and large. Decades of teaching evolutionary medicine gradually forced me to recognize that the tendency to adaptationist thinking about disease results runs deep. I began each course by warning students not to make the mistake of thinking that diseases are traits that give selective advantages. At the end of each term, despite more warnings and many examples, I received term papers proposing that schizophrenia, epilepsy, or anorexia nervosa were somehow useful.

The tendency to excess adaptationism may have an adaptationist explanation. Imagining possible functions for an object can give big advantages. Humans carve branches into digging sticks and they flake sharp stones into knives. Our words organize experience into categories based on function. Chairs come in diverse shapes and sizes, but they all are all for sitting, just as knives of many shapes are for cutting, and bags of many designs are for carrying. Because so much human cognition is organized around function, our minds are primed to pair traits with functions, even when that is not appropriate.

Educational traditions also contribute. Natural selection is taught by describing how a rare trait become common, the exemplar being the transition of the peppered moth (Biston betularia) from white to dark wings as tree trunks were increasingly stained by Victorian era-soot. The focus on how variations influence fitness is entirely appropriate; that is the process by which natural selection changes a species. However, that framework encourages viewing uncommon traits as if they are adaptations and it distracts from the vastly more common effect of natural selection to keep traits the way they are.

The second explanation for VDAA arises because many diseases have associated benefits. Those benefits do not give net advantages that can explain the trait, but they are nonetheless relevant. Cystic fibrosis alleles decrease the risk of death from cholera, but that does not mean they give a net advantage, it just means that they are selected out more slowly than they would be otherwise. The characteristics of low mood are useful in certain situations, but that does not mean that major depression is an adaptation.

Other benefits are associated with tradeoffs. Most traits reflect a compromise among multiple costs and benefits. In a stable environment, fitness tends to be highest for individuals with the average value of a trait. Although individuals with values away from the mean will have lower fitness on average, they will also experience some benefits. Individuals with more anxiety than average get protection against dangers along with larger costs from lost opportunities. Individuals with less anxiety than average benefit from risk-taking along with larger costs from harm. The wide range of most personality characteristics suggests that fitness does not vary much across the mid-range of the distribution. Individuals with traits far from the mean are, however, sometimes viewed as if they are special types selected for because they have advantages. This is usually an error. For instance, it has been repeatedly suggested that tendencies to ADHD gave selective advantages in the past, but individuals in ancestral environments who had lower-than-average ability to focus attention would be expected to have lower than average fitness in ancestral environments; the disadvantages are magnified in modern environments.

Defenses against pathogens are particularly likely to cause problems because arms races create fitness functions with steep slopes and high costs, even at the peak. So, individuals with immune systems of optimal aggressiveness are vulnerable to both infection and inflammation. Adaptive responses with flatter fitness functions also result in advantages and net disadvantages for individuals away from the mean. So, individuals with tendencies to sweat profusely get protection against overheating at the risk of dehydration, while those who sweat less than average get the opposite benefits and costs. In modern environments such costs and benefits may be shifted markedly, often creating new selection forces that shift the population mean. These observations suggest ways to consider the adaptive significance of advantages associated with disease without falling into VDAA.

In the spirit of trying to provide guidance instead of just criticism, here are three principles that can help guide attempts to find evolutionary explanations for disease vulnerability.

1. The appropriate object of explanation in evolutionary medicine is usually a shared trait that makes all members of a species vulnerable to the disease, not the disease itself.

2. Multiple explanations are usually relevant.

3. Uncommon genes or traits are usually products of mutation, migration, genetic drift, or stochasticity more generally. They are selected for only in special circumstances including:

   1. Being selected in or out. (ApoE4)

   2. Geographically separate subgroups exposed to different environments. (skin color)

   3. Effects of adaptive plasticity and facultative adaptations. (skin tanning, anxiety)

   4. Trade-offs: Individuals with trait values away from the mean are expected to have some advantages along with net disadvantages. (more than average sweating prevents heat stroke at the risk of dehydration)

   5. Balancing selection maintains genetic variations when the fitness effects of an allele vary depending on the external environment or the genomic ground. (polymorphic shell patterns in grove snails)

   6. Heterozygote advantage is a subtype of balancing selection in which the superior fitness of heterozygote individuals maintains the variation by frequency dependent selection. Heterozygote advantage is rare because the costs to homozygotes are high. (sickle cell hemoglobin)

   7. Antagonistic pleiotropy maintains alleles that cause aging because the benefits they give in youth when selection is strong are greater than the costs in late life when selection is weak. lifetime. (senescence)

   8. Morphs can persist if each has equal fitness maintained by negative frequency dependent selection of distinct separate types. (fish mating types)

   9. Bet hedging increases gene transmission in wildly varying environments at the expense of individual survival. (seed coat variation in desert plants)

In the rest of medicine, pathology is understood in the context of the functions of normal systems, and symptoms like pain, cough, and fever are recognized as useful responses. In psychiatry, both of these foundations are missing. Psychiatry has no accepted theory of how mood is gives selective advantages and how it is regulated and therefore criteria that can validate diagnoses. In their absence, mood disorders are diagnosed when the number and severity of symptoms exceeds an arbitrary threshold, without considering situations that may be arousing them. The so-called “medical model” in psychiatry is missing the foundation in understanding normal functions that the rest of medicine relies on.

Behavioral ecologists could provide that crucial missing foundation in less than a decade for a fraction of the cost of current research allocations. In the meanwhile, we must make do with a generic framework that can be summarized in three propositions. The optimal amounts and kinds of motivation, effort, and risk-taking vary depending on the situation. Individuals with capacities for special states that meet the adaptive challenges of different situations get fitness advantages. Those advantages have shaped different mood states and systems that regulate their expression. Fitness maximization requires accurately assessing what is needed in a given situation. For humans, that requires knowing an individual’s goals and life projects and expressing various aspects of mood states to the degree appropriate for the situation. Like heart or kidney failures, failures of mood control systems typically result from multiple interacting causes, rather than from discrete abnormalities that could define distinct disorders. The different kinds of mood disorders correspond not to different causes or mechanism flaws but instead to different control system failure modes.

The control systems that regulate mood are far more complex than those that maintain homeostasis. Relatively simple mechanisms monitor body temperature and regulate sweating and shivering. The systems that regulate defensive responses like cough and vomiting are more complex, but detecting foreign matter in the respiratory tract or toxins in the gut is still relatively straightforward compared to processing multiple streams of internal and external information to adjust various aspects of mood to the situation. The task is made exponentially more complex because different individuals pursue different goals in social microenvironments created by constantly shifting relationships.

Beneath the complexity, however, is an important general principle: mood disorders result from control system failures. The different failure modes of control systems should therefore correspond to recognized subtypes of mood disorders. This essay considers ten failure modes for mood control systems and how they correspond to recognized mood disorders. Some reflect the challenge of assessing situations, others from excess or deficient responses, and a few arise from fundamental control system failures.

TEN FAILURE MODES FOR MOOD CONTROL SYSTEM     Randolph Nesse 2025

1. The baseline is abnormally low or high

2. The threshold that arouses a response is abnormally low or high

3. Response intensity is excessive or deficient

4. Response duration is excessive or deficient

5. The form of the response is abnormal

6. The system responds to inappropriate cues

7. Responses are expressed in the absence of a stimulus

8. Positive feedback → vicious cycles

9. Oscillation between two extremes

10. Bistability: the response gets stuck at an extreme

Subscribe now

A control system

Ten failure modes and corresponding disorders

1. The baseline is abnormally low or high. Chronic low baseline mood is well-recognized as dysthymia; chronic high mood (hyperthymia) is more often envied than investigated or treated. Baseline mood is highly heritable, and it is as stable and as difficult to change as body weight.

2. The threshold that arouses a response is abnormally low or high. A low threshold for the intensity of cues needed to arouse depression is characteristic of borderline personality disorder but is also present in neurosis.

3. Response intensity is excessive or deficient. Excessively intense responses are closely related to—but not identical with—a low threshold of response. Deficient responses should decrease fitness but aside from studies of alexithymia and grief, they are rarely recognized as abnormal.

4. Response duration is excessive or deficient. Chronic grief has been the subject of extensive study, but responses that are too brief are rarely described.

5. The form of the response is abnormal, for instance when eating and motivation are dramatically lowered without any subjective low mood. Calling such responses “desynchronized” reflects tacit creationism that expects uniform responses and it neglects work by Matthew Keller finding that different aspects of low mood are expressed in different situations in patterns that appear adaptive.

6. The system responds to inappropriate cues. A depressive response to cues that are not actually germane for an individual is characteristic of neurosis, where sensitized systems constantly scan for hints of rejection or criticism, and express many false alarms. However, other conditions also fit here, such as depression in response to a public figure’s death that will not influence a person’s life.

7. Responses are expressed in the absence of a stimulus. Apparently autonomous episodes of depression or mania constitute a failure mode that differs markedly from most others discussed here. In this mode a defective control system expresses the emotion in the absence of any identifiable relevant situation. However, the absence of a life event alone is not sufficient to define this mode because inflammation and unconscious processes can be covert provocations.

8. Positive feedback loops create vicious cycles that make mood disorders worse. Depression escalates and endures when it results in avoiding friends and exercise and not pursuing reachable goals. Severe depression can also change the brain in ways that make future episodes more likely. Mania is often a product of a similar vicious cycle in which success, or the perception of success, increases motivation and risk-taking, causing greater apparent success in a toxic spiral. Other episodes of mania can be autonomous from life experiences.

9. Oscillation is a failure mode that associated with systems that have very high gain. Modern thermostats use anticipation mechanisms to minimize swings, activating or deactivating a furnace or air conditioner just before the set point is reached. The apparently inexplicable low mood many people experience after a major success may protect against manic escalation. Deficiencies of this damping mechanism may cause the moderate mood swings in cyclothymia. The extreme mood swings in bipolar disorder induce taking large risks in pursuit of grandiose goals. Manic episodes often end abruptly, as if a regulatory mechanism has abruptly suppressed motivation. Such extreme swings are a characteristic of failure modes resulting from excessive gain in the system. Research on subtypes of manic-depressive illness and cyclothymia has steadily expanded recognition that mild mood swings are very common suggesting there may be tradeoffs that maintain somewhat high gain in the system.

10. Bistability, the tendency for a system to get stuck at one or the other extreme, is also characteristic of systems with very high gain. Its manifestation in bipolar disorder supports the potential importance of this failure mode.

Why recognizing mood disorders as control systems failure modes is important

First, and most importantly, it recognizes aversive emotions as defensive responses similar to pain, fever, and fatigue. That recognition should encourage careful examination of the patient’s motivational structure: values, goals, plans, and how things are going for goal pursuits in each area of life. The S.O.C.I.A.L. list of resources is a useful framework for guiding such investigations; it reveals relevant life situations and equally often, their irrelevance.

Second, framing mood disorders as modes of control system failure promotes analysis at the system level. This approach helps reduce tendencies to assume that different disorders must arise from different specific abnormalities.

Third, different failure modes are likely to correspond to different neural and psychological mechanisms. Using failure mode categories to classify research subjects may yield stronger results.

Fourth, and the topic for a future essay, are the intrinsic vulnerabilities of control systems. Optimal function requires some false alarms, as illustrated by the smoke detector principle. Operation in novel environments often causes control system dysfunctions. And the nature of organically complex systems that makes them robust also leaves them vulnerable to failures, especially from self-adjusting response thresholds that can initiate positive feedback loops creating syndromes including chronic pain.

Finally, analysis in a control system framework emphasizes that organic systems have structures of complexity that are fundamentally different from those of designed machines. Unlike designed machines, which have discrete parts with specific functions, organic control systems consist of interwoven, highly networked components. This organization make organic systems remarkable robustness, but also makes them vulnerable to failures that cannot be traced to abnormalities in any single component.

For a printable PDF, see https://www.randolphnesse.com/nessays

When William Bateson coined the term “epistasis” in 1909, he imagined one gene standing (“epi”) on top of another (“stasis”) and thereby blocking its effects. If he had instead simply called such phenomena “non-additive genetic interactions” vast confusion would have been prevented, but we are stuck with the term. Epistasis now refers more broadly to any instance in which the effect of one mutation on a phenotype is changed by the presence of another mutation. I use the term “gene” instead of allele because our model is haploid, and I call variants “mutations” for simplicity, even though the variations can have other origins.

Quantifying epistasis requires four values:

• the phenotype (PT) when all genes are at wild type (WT),

• the change in the PT value when only gene A is added, or the PT value when all other genes are at WT (PTA)

• the change in the PT value when only gene B is added, or the PT value when all other genes are at WT (PTB), and

• the PT when both A and B are present (PTA&B).

The first step in calculating epistasis is to find the Additive Expectation (AE).

AE is the PT value predicted if there are no interaction effects so that the sum of the independent effects of gene A and gene B plus the WT gives the value of PTA&B.
AE is calculated by adding (PTA-WT) and gene B (PTB-WT) to WT.

• AE = (PTA - WT) + (PTB - WT) + WT, simplifies to

• AE = PTA + PTB - WT.

In the above figure, WT is 10, PTA is 4 above WT, PTB is 2 above WT, so the AE is 10+4+2=16.

Epistasis is the difference between PTA&B and the AE.

• Epistasis = PTA&B – AE which is the same as

• Epistasis = PTA&B – (PTA + PTB + WT) which is the same as

• Epistasis = PTA&B – PTA - PTB – WT.

In the above figure, the gene combination A&Bv1 gives a phenotype of 19, which is 3 more than the WT of 16, so the epistasis is +3. For gene combination A&Bv2 the phenotype is 14, 2 less than the WT, so the epistasis is negative at -2.

The additional figures below illustrate AE and epistasis when both mutations reduce the PT, and when one mutation increases the PT and the other reduces it.

I hope these illustrations and formulas help to clarify what epistasis is and how to calculate it, and would be glad to hear from those who find them useful and those who have suggestions.

A PDF version is available at this link and more is at RandolphNesse.com

When aspiring evolutionary psychiatry researchers ask for suggestions for good research topics, I have often suggested that they study the foraging patterns of people with ADHD to see if their patch staying times are too short. I was delighted to find a recently published study that used a design very close to the one I advised students to pursue!

Barack, D. L., Ludwig, V. U., Parodi, F., Ahmed, N., Brannon, E. M., Ramakrishnan, A., & Platt, M. L. (2024). Attention deficits linked with proclivity to explore while foraging. Proceedings of the Royal Society B: Biological Sciences, 291(2017), 20222584. https://doi.org/10.1098/rspb.2022.2584

The design and methods are appropriate. The study was preregistered, and an adequate representative sample of 467 subjects was recruited and studied online using a computerized foraging task and questionnaires about ADHD. The subjects were instructed to get as many berries as possible by deciding, at every instant, whether to stay in a patch as the rate of berry accumulation gradually slows, or move to a new patch, with a travel time delay. Information about which questionnaires were administered when was not provided, despite the risk of demand characteristics influencing the results.

Only 24 subjects reported a previously receiving an ADHD diagnosis, but about half of the 467 subjects had scores on ADHD-associated items high enough to screen positive for ADHD, making the validity of the categorization suspect.

Subjects stayed much longer in each patch than would be optimal. When the delay was 1 second, they stayed an average of 20 seconds, much longer than the optimal durationof 12 seconds. Increasing the delay to 5 seconds increased staying-time to 26 seconds, more than twice as long as the optimum of 12 seconds. While many studies find longer than optimal staying-times, the suboptimality in this study might relate to the very short optimal staying times and the relatively short duration of the trial for each subject.

In the main multivariate analysis, it was a surprise to see that patch staying-time was not associated with the ADHD screening score; instead it was strongly associated with the overall rate of reward. I found this hard to understand given that the overall excessive staying-time for most individuals meant that any tendency to a shorter time would increase the number of berries gained. Another analysis found shorter staying times for those who screened positive for ADHD.

The below illustration shows that individuals who screened positive for ADHD had greater average reward rates than those who screened negative, but both were far from optimal.

These findings are used to support the conclusion that "Consistent with our reported findings, we speculate that ADHD serves as an adaptive specialization for foraging [68–71], thus explaining its widespread prevalence and continued persistence in the human population." (p.9) The last sentence in the abstract phrases the same conclusion a bit differently: " Our findings suggest that ADHD attributes may confer foraging advantages in some environments and invite the possibility that this condition may reflect an adaptation favouring exploration over exploitation."

Leaving aside the findings themselves for now, the hypothesis that ADHD is an adaptation is hard to reconcile with evolutionary theory. Natural selection tends to shape traits to mean values for a species that maximize individual inclusive fitness, so values away from the mean tend to be associated with lower fitness. Of course, environments vary, so fitness may be about the same across a wide range of trait values, especially across multiple generations. Also, across many more generations in a new environment a variation may give a selective advantage that justifies calling it an adaptation. However, these possibilities do not support the suggestion that ADHD is an adaptation that gives fitness advantages.

The logic of this study does, however, make it clear that ADHD associated characteristics have important adaptive consequences. Several viable categories of hypotheses receive support in animal studies. The first is that selection has shaped facultative adaptations that adjust an individual's foraging strategies as a function of experience. This is widely confirmed. The second is that different foraging environments will shape different foraging parameters for different species and different subgroups within a species. This too is confirmed. Both ideas are, however, quite different from the hypothesis that ADHD is an adaptation.

Like most traits, faster or slower shifts in attention have trade-offs, so individuals with trait values away from the mean are expected to get some advantages, along with net disadvantages. Individuals with ADHD tendencies may well benefit from more exploration than other individuals, but that is different from saying that ADHD itself is an adaptation.

In conclusion, the impetus behind the study is great! Variations in ADHD associated characteristics have important influences on adaptation and fitness and experiments based on foraging theory are a great way to study them. But ADHD is not an adaptation.

The greatest advance in the history of medicine was recognition that bodies are not products of supernatural creation, they are machines assembled entirely from material substances. Now, however, the metaphor of body as machine is an obstacle to progress. To understand the body as a body, we need to blow the metaphor away. Why will be clear only if we first acknowledge its origins and benefits.

The life force: a seductive idea

When death transforms a living breathing body into a cold corpse, it seems obvious that something has gone missing. This mysterious substance has had many names. In 1907 the French philosopher Henri Bergson used the term élan vital, often translated as “vital force.” The idea has endured since the time of the ancient Greeks. Aristotle was preoccupied by the difference between inanimate and living things, perhaps because was the son of a physician to the king of Macedon, the most influential Greek kingdom. The title of his book, De Anima, is usually translated as “On the Soul,” but he viewed the soul as the essence of living organisms, a life force.

A hundred years later, the great Greek physician Galen taught that the special life substance was pnuema, taken in via the lungs and spread throughout the body. He had evidence: when breathing stopped, life ended. For over a thousand years most people in the West went along with the idea that living things have some special life substance that makes them different from inanimate objects. A few pesky early Greek materialists including Democritus and Epicurus thought no such special substance was needed to explain life. However, for most people, for most of human history, it has seemed obvious that living bodies are inhabited by some special life force. It persists today in talk of energy fields. At its root, this “vitalism” is the idea that some special nonmaterial energy or substance is required to explain life.

Leonardo da Vinci: Great Lady

The origins of the machine metaphor

Once dissections became common and anatomical drawings by Leonardo da Vinci and Andreas Vesalius began to circulate in the early 1500s, it became pretty obvious that bones and muscles were just fancy systems of levers, ropes and pulleys. Nothing mysterious, just things. But it was not until the early 1600’s that the French philosopher René Descartes argued for replacing vitalism with scientific materialism. Most people think, “Oh yeah, Descartes. He is the cogito ergo sum guy who created the big nuisance of the mind-body problem, isn’t he?” Yes, that’s him. But give the guy a break! Convincing the world that the body is a machine was a very big deal. He got into plenty of trouble for proposing the radical concept of the body as a machine. Suggesting that the mind too was a machine would have been touching the third rail!

Renee Descartes

As the industrial revolution transformed society, the metaphor of body as machine became increasingly influential. By the start of the 20^th century, the idea dominated thinking in biology and medicine, probably because it is so useful. It has improved our lives by encouraging detailed analysis of the body’s mechanisms at all levels, from the details of anatomy, to understanding how hormones like insulin regulate chemicals like glucose. It encouraged reductionism, the idea that everything large could be explained by analysis of smaller things. We are now down to genes, molecules, and atomic forces. What an extraordinary bounty we have reaped from a metaphor! The metaphor of body as a machine provided a ladder that allowed biology to bring phenomena up from a dark pit of mysterious forces into the light where organic mechanisms can be analyzed as if they are machines.

The body is not a machine

However, the body is not a machine. Machines are products of design, bodies are products of natural selection, and that makes them different in fundamental ways. The organic complexity of bodily mechanisms is qualitatively different from the mechanical complexities of machines. Machines have discrete parts with specific functions connected to each other in straightforward ways. Bodies have parts that may have blurry boundaries and many functions and the parts are often connected to each other in ways hard for human minds to fathom. Bodies and machines fail for different reasons. Engineers can start from scratch if they need to in order to fix weak spot in the design of a machine. If only our human spine could be redesigned from scratch! Its limits and compromises are the source of vast pain, but natural selection can’t start fresh, so we are stuck with a substandard design that can be improved only by small changes. The Table illustrates the substantial differences between machines and bodies. These differences are, however, often ignored, in large part thanks to the power of the metaphor, and the fear that setting it aside will lead to the resurgence of vitalism.

(See our 2025 article in PNAS Nexus for an updated table comparing bodies to machines)

For many scientists, hearing the phrase “the body is not a machine“ is a cue that arouses an almost automatic attack on vitalism. They assume that any derogation of the machine metaphor is an attempt to sneak in vitalism in new vestments. Their wariness is understandable. Naïve talk about the life force or energy fields has to be weeded out of medicine as steadily as crabgrass from a lawn. It is not a big problem. What is a big problem is the metaphor of body as machine; it is as pervasive and pernicious now as vitalism was in the Middle Ages. OK, that is an exaggeration. The metaphor is not AS bad as vitalism. It does, however, distort thinking in ways that slows progress.

One powerful example is how we teach biochemistry and physiology. We describe systems using idealized diagrams with boxes and arrows. For instance, every medical student memorizes (then forgets) the chain of chemical interactions that make blood clot. This knowledge is essential for understanding clotting disorders, but the diagram is distant from the reality.

Current research often relies on tacit models of body systems as if they were designed. (See my essays on “tacit creationism,” the pervasive tendency to view living systems as if they are products of design, without any reference to supernatural forces.) A multi-billion dollar effort is trying to discover the “wiring diagram of the brain.” But is there a master wiring diagram? The White House Brain Initiative will be more effective if it recognizes that there is no one normal genome, no one normal brain, and no one wiring diagram. Similarly huge efforts are mounted to discover the functions of each location in the brain. The amygdala, a tiny almond shaped area deep in the sides of our brains, has often been described as the locus of fear learning. Yes, if the amygdala is damaged, fear learning suffers. However, many other regions are involved in regulating fear, and the amygdala serves many functions aside from fear learning, including social responses, self-control, aggression, and learning to get positive rewards.

The metaphor of body as machine is a serious error with major costs. In psychiatry, thinking about the mind as a machine has led to a debacle about diagnosis. Many neuroscientists want to abandon the standard system because they cannot find specific brain abnormalities for any of the major disorders. They are sure that for every disease there is some findable broken part. If only. Many mental disorders are, like heart failure, failures of systems with multiple causes and diverse symptoms.

A soma is a soma is a soma

Metaphors are robust. Unless something better is at hand, criticizing the metaphor of the body as machine could be a waste of energy. An alternative metaphor would be ideal, but there is no other metaphor for the body that does not distort reality. Instead of looking for alternative metaphors, we must acknowledge the reality of organic complexity, with its blurry boundaries between bird’s nests of mechanisms. Instead of a metaphor, we must recognize the body as a soma shaped by selection. Soma just means body, so that only helps a little, until we turn for help to the poet Gertrude Stein. Remember her famous, “A rose is a rose is a rose.” She meant it. Eschew metaphor, she said; see the rose as a rose, just the rose itself.

A soma is a soma is a soma, shaped by natural selection. If we can wrench ourselves away from metaphor and see the body as an evolved soma, we can put aside the debates that arise from assuming that its parts have nice crisp boundaries and specific functions. We can avoid further centuries of debates about exactly how many basic emotions they are. Our bodies have parts with blurry boundaries, multiple functions, unimaginably complex connections, and no designer’s purpose underlying the whole system. We teach students distorted versions of biological systems that are simple enough to memorize, and to test! But our diagrams are idealized abstractions that misrepresent reality in fundamental ways. Many organic molecules influence many other kinds of molecules, not just the one to the left and the one to the right in an idealized diagram.

We are on the verge of a transformation. Giving up our idealized view body in the machine will be challenging, but worth it as we accept the reality that a soma is a soma is a soma.

Leave a comment

Further reading

Nesse, Randolph M, and George Christopher Williams. Why We Get Sick: The New Science of Darwinian Medicine. Vintage, 1996.

Williams, George C. The Pony Fish’s Glow : And Other Clues to Plan and Purpose in Nature. 1st ed. New York, NY: Basic Books, 1997.

Balleine, B. W., & Killcross, S. (2006). Parallel incentive processing: an integrated view of amygdala function. Trends in Neurosciences, 29(5), 272–279. http://doi.org/10.1016/j.tins.2006.03.002

Nesse, Randolph M: Recognizing that the body is not a machine. EDGE annual question essay, 2009. https://edge.org/response-detail/1136.

See this link for more recent work on tacit creationism.

1. What is evolutionary medicine?

How I look forward to the day when no one needs to ask! Evolutionary medicine, sometimes called Darwinian medicine, is the field at the intersection of evolution and medicine. Evolution AND Medicine is a better descriptor, but that has not caught on. The field uses the basic science of evolutionary biology to find ways to prevent and treat disease, and it uses studies of disease to advance basic evolutionary biology(2).

2. What isn’t evolutionary medicine?

Evolutionary medicine is not radical or alternative. It is not a special kind of medical practice. It does not advocate any particular kind of diet, exercise, or treatment. It does not make direct clinical recommendations based on theory alone (3). If your doctor claims to practice evolutionary medicine, be sure that treatment recommendations are based on controlled scientific studies; if they come directly from theory, find a better doctor. Evolutionary medicine is standard mainstream medical science, appropriately cautious about making clinical recommendations.

3. How does evolutionary medicine transform our understanding of disease?

Evolutionary medicine poses a fundamentally new kind of question about disease. Instead of only asking how bodies work and why some people get sick, evolutionary medicine also asks why natural selection has left all of us with traits that make us vulnerable to disease (1). Why do we have wisdom teeth, narrow coronary arteries, a narrow birth canal, and a food passage that crosses the windpipe? Evolution explains why we have traits that leave us vulnerable to disease, as well as why so many other aspects of the body work so well. For instance, the usual question about back pain is why it afflicts some individuals. Evolutionary medicine also asks why back problems have been a problem for all hominid species since they first walked on two legs.

4. Is evolutionary mainly about diseases caused by our modern environments ?

Most chronic disease results because our bodies are poorly prepared to stay healthy in modern environments(4), but this is only one of several evolutionary explanations for why we get sick(5). Vulnerability to infections persists because pathogens evolve so quickly. Not all deleterious mutations can be eliminated because there are limits to what natural selection can do. Other vulnerabilities result from inevitable tradeoffs, including the big one of maximizing reproduction even when that harms health. Also, selection has shaped many defenses, such as inflammation, whose utility comes at a substantial cost in tissue damage and suffering. These are not alternatives to standard medical explanations, they are complementary deeper explanations.

5. Is evolutionary medicine only about vulnerabilities?

No, questions about vulnerability are only one part of evolutionary medicine. Some scientists use DNA evidence to trace the evolutionary history of humans and pathogens. Some use evolution to discover new ways to prevent resistance to antibiotics and cancer chemotherapy drugs. Others study why selection cannot remove some genetic variations that cause disease. Evolutionary medicine includes all aspects of evolutionary biology applied to all problems in medicine and public health.

6. Evolutionary medicine shows how diseases are useful, right?

Wrong. Diseases are not adaptations shaped by selection. There is nothing useful about pneumonia, schizophrenia, epilepsy, or cancer. Trying to understand diseases as if they are adaptations is a mistake, one that is unfortunately as common as it is serious (6). However, many symptoms of disease, such as pain, fever, vomiting, cough, and fatigue, are adaptations. The systems that regulate such defenses are, for good evolutionary reasons, prone to failures that cause chronic pain, anxiety disorders, and many other diseases.

7. Is evolutionary medicine useful?

It sure is! Some applications are relatively direct— doctors who understand how selection shaped regulation mechanisms can make better decisions about when it is safe to use drugs to block fever and cough(2,3). However, like genetics and microbiology, evolution is a basic science that mainly provides new understanding that lead to new treatments(1,7,8). Evolutionary models that analyze the evolutionary competition between cells in a malignancy have led to new chemotherapy strategies that greatly extend the lives of mice with cancer(9). Models of antibiotic resistance suggest that “take every pill in the bottle” may be bad advice(10). We must find out what in our modern environment is causing the new epidemic of autoimmune diseases like diabetes, multiple sclerosis, and Crohn’s disease(11).

8. Don’t most doctors already know all about evolution?

No, most doctors would likely fail the midterm in an introductory evolution class(13,14). Even medical leaders often have gross misconceptions. A famous geneticist once suggested at a big Darwin meeting that selection keeps mutations happening for the good of the species, an idea has been recognized as a mistake for 50 years. The belief that long association with a host makes viruses milder remains common among physicians, decades after Paul Ewald and others pointed out that selection maximizes transmission of pathogens even if that kills the host. Engineers learn the principles of thermodynamics so they don’t make mistakes like trying to build perpetual motion machines. Doctors never learn the principles of evolutionary biology, so major misconceptions persist in medicine(15).

9. Why don’t medical schools teach evolution the same way they teach other sciences?

Most medical schools have no evolutionary biologists on faculty, few doctors who know much about evolution, and some who think bodies are products of intelligent design. Even the core idea that evolutionary explanations are needed to complement descriptions of mechanisms is unfamiliar to most doctors. Some display their naiveté of methods for testing evolutionary hypotheses by spouting the phrase “just so stories,” as if that is a devastating critique. Some actually teach “just so stories.” That evolution in the medical curriculum remains in the 19th century is shameful.

10. How long will it take for medicine and public health to make full use of evolutionary biology?

Progress is now rapid(16). The International Society for Evolution, Medicine and Public Health has created a network of evolutionarily sophisticated researchers, educators, and clinicians, and new education resources, including EvMedEd.org. Several new textbooks are available (8,17,18) along with a new journal, Evolution, Medicine, & Public Health, and The Evolution and Medicine Review. The evolution questions recently added to medical school entrance examinations are spurring enrollment in evolution courses. Students who have taken an undergraduate evolutionary medicine course are arriving at medical school full of new questions. They will grow up to be medical school deans. When they do, evolutionary biology will finally be recognized widely as an essential basic science for medicine.

Literature Cited

(More recent resources are available at isemph.org and at RandolphNesse.com)

1. Nesse RM, Williams GC. Why we get sick : the new science of Darwinian medicine. 1st Vintage Books. New York: Vintage Books; 1994. xi, 290 p.

2. Nesse RM, Stearns SC. The great opportunity: Evolutionary applications to medicine and public health. Evol Appl. 2008;1(1):28–48.

3. Nesse RM. Ten questions for evolutionary studies of disease vulnerability. Evol Appl. 2011;4(2):264–77.

4. Gluckman PD, Hanson M. Mismatch: why our world no longer fits our bodies. New York: Oxford University Press, USA; 2006.

5. Nesse RM. Maladaptation and natural selection. Q Rev Biol. 2005 Mar;80(1):62–70.

6. Nesse RM. Ten questions for evolutionary studies of disease vulnerability. Evol Appl. 2011;4(2):264–77.

7. Stearns SC. Evolutionary medicine: its scope, interest and potential. Proc Biol Sci. 2012 Nov 7;279(1746):4305–21.

8. Stearns SC, Medzhitov R. Evolutionary medicine. Sunderland, MA: Sinauer Associates, Inc., Publishers; 2016.

9. Gatenby RA, Silva AS, Gillies RJ, Frieden BR. Adaptive therapy. Cancer Res. 2009 [cited 2016 Jun 3];69(11):4894–903.

10. Read AF, Day T, Huijben S. The evolution of drug resistance and the curious orthodoxy of aggressive chemotherapy. Proc Natl Acad Sci. 2011;108(Supplement 2):10871–7.

11. Blaser MJ. Missing microbes: how the overuse of antibiotics is fueling our modern plagues Macmillan; 2014.

12. Nesse RM. Natural Selection and the Regulation of Defenses: A Signal Detection Analysis of the Smoke Detector Principle. Evol Hum Behav. 2005;26:88–105.

13. Nesse RM, Schiffman JD. Evolutionary Biology in the Medical School Curriculum. Bioscience. 2003;53(6):585–7.

14. Hidaka BH, Asghar A, Aktipis CA, Nesse RM, Wolpaw TM, Skursky NK, et al. The status of evolutionary medicine education in North American medical schools. BMC Med Educ. 2015 [cited 2016 Feb 1];15(1):1.

15. Antolin MF, Jenkins KP, Bergstrom CT, Crespi BJ, De S, Hancock A, et al. Evolution and medicine in undergraduate education: a prescription for all biology students. Evolution. 2012 Jun;66(6):1991–2006.

16. Nesse RM, Bergstrom CT, Ellison PT, Flier JS, Gluckman P, Govindaraju DR, et al. Making evolutionary biology a basic science for medicine. Proc Natl Acad Sci USA. 2010 Jan 26;107 Suppl 1:1800–7.

17. Perlman RL. Evolution and medicine. First edition. Oxford, United Kingdom: Oxford University Press; 2013. 162 p.

18. Gluckman P, Beedle A, Hanson M. Principles of Evolutionary Medicine. Oxford, UK: Oxford University Press; 2009.19. Kruger DJ, Nesse RM. An Evolutionary Life-History Framework for Understanding Sex Differences in Human Mortality Rates. Hum Nat. 2006;17(1):74–97.

20. Nesse RM. Tinbergen’s four questions, organized: a response to Bateson and Laland. Trends Ecol Evol. 2013 Dec;28(12):681–2.

By constructive engagement, I mean people trying as hard as they can to express their own ideas clearly, to understand other people’s beliefs and their reasons for those beliefs, to understand the exact differences between the beliefs, and to specify and search out information that would resolve any differences. Far more common are two patterns of unconstructive engagement, which can be caricatured as the mutual admiration society and the war of the clans.

In the mutual admiration society, members flatter and congratulate each other, falling all over their own ideas in order to show that there are not any real, or at least any serious disagreements, and that the other people’s ideas, and the people themselves, are wonderful and superior to others. Good ideas and areas of agreement are emphasized, while errors, contradictions, and disagreements are ignored, excused, or actively suppressed. This is the norm in most public settings, including academia.

In the war of the clans, participants perceive and describe the beliefs of others as negatively as possible, and they attack whatever is most vulnerable, including the people, their abilities, motives, and moral standing. Good ideas and areas of agreement are ignored or actively suppressed, while errors, contradictions, and disagreements are emphasized. The participants in clan wars usually seek allies who will join them to support a shared ideology that uncritically accepts and glorifies the beliefs of the in-group, and that rejects and ridicules the beliefs and members of the out-group.

These modes of engagement are intimately related. It is dangerous to attempt to try to tell a friend or colleague that you think he or she is wrong about something because the expectation for mutual support in the in-group is so strong that any deviation, even the most careful and respectful attempt to explore areas of disagreement, is experienced as a defection, as a social attack that indicates opposition. Most people want friends and they don’t want enemies, so they mostly say what others want to hear. Muddled ideas and vagueness triumph.

This is why politics is so difficult despite being ubiquitous, and perhaps why its difficulty may have led to the rapid evolution of the human brain. The politician must simultaneously convince people that he or she will advance the interests of the in-group, usually by opposing an out-group, while ensuring that the out-group is small and powerless enough to not be a threat. Coalitions are built on common interests, but if the interests are too broad, people will not identify with the group, while if too narrow, the group will be small. The solution is to say little of substance, appearing always to support the agendas of those who are listening, while attacking ill-defined outsiders. To speak clearly about specific beliefs and plans is suicidal for a politician. To speak clearly with our friends and colleagues about areas of disagreement is dangerous for any of us.

What is evolutionary about all of this? Our ancestors have lived in complex social settings for well over a million years. People who have a tendency to engage in mutual admiration behavior with in-group members, who conform to, believe in, and advocate and work for the ideology of the in-group against the interests of other groups, have almost certainly have had a selective advantage over people who tend to be more objective. Subjectivity and the emotions that fuel it are not flaws, but adaptations.

In academia, we highly value constructive engagement, but we don’t do it very often. Far more often, we politely engage in mutual admiration, letting crucial disagreements persist for the sake of social alliances. Less often, but even more unconstructively, we engage in clan wars, where opposing sides define group boundaries and then distort and derogate the out-group’s members and their ideas and beliefs. Examples of such wars are endless: post-modernists vs. scientists, behaviorists vs. Freudians, Marxists vs. free-marketers, social scientists vs. biological scientists, evolutionists vs. creationists, geneticists vs. whole organism biologists, adaptationists vs. Gouldians, conservatives vs. liberals, pro-lifers vs. abortion rights advocates, theoreticians vs. experimentalists, and all the proponents of various theories of literary criticism against each other. These examples are not anomalies, they are what humans do, and how human societies are organized. Worse yet, people have an intrinsic tendency to attend to a person who is passionately dedicated to a simple, vivid, unusual position. So leaders tend to polarize. Mass media bring instant attention to the most extreme positions, amplifying the phenomenon.

Constructive engagement always involves risk and often requires sacrifice. Even to advocate it may be seen as an indication that one can not be counted on for unswerving loyalty to the in-group. Constructive engagement is rare for good reasons, but it is nonetheless wonderful.